King and Pawns Endgame Stalemate Trap

Here's a useful one I've come across in my studies:

White deliberately stalemates himself to avoid the losing pawn endgame. Note that if it was and A pawn it wouldn't work because black could then play Qh8# as the final move (unless blacks King was in the way).

The order of the moves is not unique. Better is a version like this one:

How is that different exactly?

Black has an e pawn instead of a b pawn. Your example is nice, but White's move order makes no difference. In the problem shown by @Arisktotle, the move order has to be precise in order to account for the possibility of underpromotion.

Oh I see! If white pushes the G pawn first, blacks knight promotion guards G2

And of course if white pushes the h pawn sooner, black checkmate with NG2, nice!

And in thos case the order wouldn't matter black would win with the knight if the king is close enough.

The last diagram already fails to 1. Kh4 Qe1+2. g3 Qa1 0-1. There are many studies possible with this self-stalemate pattern and many already exist. Another one is in post #9.